Download Design, fabrication and performance of a 1.29 T Bi

INSTITUTE OF PHYSICS PUBLISHING
SUPERCONDUCTOR SCIENCE AND TECHNOLOGY
Supercond. Sci. Technol. 14 (2001) 433–443
www.iop.org/Journals/su
PII: S0953-2048(01)22405-1
Design, fabrication and performance of a
1.29 T Bi-2223 magnet
A B Sneary1 , C M Friend2 and D P Hampshire1
1
2
University of Durham, Department of Physics, South Road, Durham DH1 3LE, UK
BICC General Superconductors, Oak Road, Wrexham LL13 9XP, UK
Received 16 February 2001
Abstract
The design and fabrication procedure of a laboratory-scale Bi-2223 tape
superconducting magnet with a bore of 40 mm and a maximum field of
1.29 T at 4.2 K is presented. The magnet comprises six resin impregnated
double-wound pancakes of bore diameter 40 mm fabricated via the
react-and-wind route. Critical current density (Jc ) measurements have been
made as a function of magnetic field, angle and strain at 4.2 K and 77 K on
short samples. In zero field, the critical current density for the
superconducting cross-sectional area of the tape was 8.3 × 104 A cm−2
(4.2 K) and 1.18 × 104 A cm−2 (77 K). The electric field–current density
characteristics of all the components of the coils when individually
energized or with the whole magnet energized have been measured.
Comparison between short sample measurements and performance of the
magnet show that minimal additional damage occurred beyond the ∼20%
that was produced by the bending strain during the wind-and-react
fabrication procedure and the ∼10% variation of the long length Jc of the
tape. Sufficient detail is provided for the non-specialist to assess both the
use of potential brittle superconducting tapes for magnet technology and to
construct a laboratory-scale magnet.
1. Introduction
Since the discovery of Bi-based high-temperature superconducting ceramics [1, 2], an enormous effort has gone into improving the critical current density (Jc ) of these materials. In
short samples, Jc of 6.9×104 A cm−2 [3] and 7.3×104 A cm−2
[4] has been achieved in zero-field at 77 K. High-field applications rely on the use of high-performance tapes of kilometre
lengths. In the Bi-2223 system, this level of performance is
now being realized with Jc of 1.2 × 104 A cm−2 reported
over 1260 m [5], 1.8 × 104 A cm−2 over 250 m [6] and over
2 × 104 A cm−2 achieved in 1200 m [7], 1000 m [8] and 400 m
[9] lengths at 77 K in self-fields. It is expected that Jc will
increase yet further, perhaps by an order of magnitude [10], as
our understanding of tape processing improves.
The high current-carrying capacity of Bi-2223 up to 30 K
[7, 11, 12] enables the production of compact cryogen-free
magnet systems. Cryogen-free systems are attractive because:
(i) easy operation is facilitated since no cryogen is required;
(ii) cooling efficiency at 20 K is approximately five times
higher than at 4 K making refrigeration more economical;
(iii) thermal stability and fast magnet sweep rates are possible
because the specific heat of the tapes is about two orders
0953-2048/01/070433+11$30.00
© 2001 IOP Publishing Ltd
of magnitude higher at 20 K than at 4 K. Cryogen-free Bi2223 systems have been fabricated using stacked pancake coils
producing fields of 7.25 T [13] and 7 T [14] at around 20 K.
The high upper critical field of Bi-2223 also facilitates the
production of high-field inserts. A record field of 24 T at
4.2 K [7] was achieved from a Bi-2223 insert in a 22.54 T
background field, the same magnet producing 2.3 T at 4.2 K in
a self-field. A magnet producing 3.2 T in a self-field and 1 T
at 4.2 K in a background field of 20 T has been fabricated [5].
Inevitably the enormous effort producing large-scale Bi-2223
magnets and inserts has led to the proprietary use of materials
and commercial knowledge.
Increasing opportunities for fabricating laboratory-scale
magnets in the 1–5 T range operating at temperatures below
30 K will follow the improvements in Jc of Bi-2223 tapes.
Laboratory-scale magnets have been produced using both
the ‘react-and-wind’ (R + W) and ‘wind-and-react’ (W + R)
processes. Using R + W, 1 T has been generated [15] from
eight R + W double-wound pancakes at 4.2 K. At 20 K, 1.1 T
was achieved from 17 R + W double-wound pancakes [16].
Using W + R, a Bi-2223 coil with a 15 mm bore generated a
field of 1.34 T at 4.2 K [17] and Bi-2212 inserts have produced
2.6 T in a self-field and 1.08 T in a background of 20 T at
Printed in the UK
433
A B Sneary et al
Table 1. The main parameters and dimensions of the Bi-2223 tapes.
Material
Sheath
Thickness (mm)
Width (mm)
Number of filaments
Fill factor (%)
Length of tape (m)
Ic of sections along tape length at
77 K, zero magnetic field (A)
Tape 1
Tape 2
Bi-2223
Silver alloy Ni/Mg
0.33
3.3
37
28
120
Bi-2223
Silver alloy Ni/Mg
0.29
3.5
37
28
70
35–38
33–35
1.4
40
1.2
JC(104 A.cm-2)
30
0.8
20
0.6
IC(A)
Straight thermal cycles
Straight mechanical cycles
Bent mechanical cycles
1.0
0.4
10
0.2
0.0
0
1
2
3
4
5
6
Number of cycles
7
8
9
0
10
Figure 1. The critical current density (Jc ) of thermally and mechanically cycled 100 mm samples of Bi-2223 tape at 77 K. The open
symbols () are for thermally cycling alone. The closed symbols are for a tape that is bent (•) and Jc measured, and then straightened ()
and remeasured for 10 cycles.
4.2 K [18]. This paper includes an outline of the fabrication of
a laboratory-scale magnet that was produced without the use
of specialist equipment (e.g. coil winding machines).
In section 2, Jc data for short samples of Bi-2223 tape
are presented and used to assess the effects of magnetic field,
strain and anisotropy. In section 3, the design and fabrication
procedure for the Bi-2223 magnet system consisting of six
double wound pancakes is presented. Since the design
parameters of any laboratory-scale magnet are specific to the
application including magnitude of field, field homogeneity,
reinforcement of the structure and bore size, alternative
available fabrication options are also discussed. In section 4,
Jc data for the magnet and all constituent pancakes at 4.2 K
and 77 K are presented. Axial field profiles of the magnet,
measured using a Hall probe, are compared with calculations.
The electric field–current density (E–J ) characteristics of
all pancakes were measured both with each coil (i.e. double
pancake) individually energized and with the magnet energized
to assess the reliability of the fabrication route. In section 5,
the performance of the coils and pancakes is compared
with calculations which together demonstrate that the magnet
presented produced a field close to short-sample performance.
The discussion in section 6 considers the data presented and
suggests a series of experiments required to efficiently assess
the suitability of a particular tape for a particular application.
In this paper, sufficient detail is included so that the non434
specialist can assess both the performance of available brittle
superconducting tapes for magnet technology and then design
and fabricate a R + W laboratory-scale magnet using pancake
coils with equipment available in any well-founded laboratory.
2. Measurements on a short sample of a BiSCCO
tape
Once the basic design parameters of magnet bore size and
dimensional restrictions are determined (cf section 4), the field
strength of modern magnets is almost entirely determined by
the performance of the superconducting wire or tape [19].
In our magnet, system restrictions include a minimum bore
diameter of 40 mm, maximum outer diameter of 82 mm and
use of the R+W conductor. In this section, short samples of the
BiSCCO tape used to fabricate the magnet are characterized.
The tapes used were 37 filament Bi-2223 alloy sheathed
tape fabricated via the powder-in-tube (PIT) route with Jc of
8.3 × 104 A cm−2 (4.2 K, 0 T) across the superconducting
cross section. The parameters of the two constituent tapes are
presented in table 1. In Bi-2223, it is known that Jc is strongly
dependent on the magnitude of the field [7, 11, 12], angle of
the applied field [20–26] and strain [27–30].
Figure 1 shows the effects of thermal and mechanical
cycling on the critical current density of tape 1 at 77 K and
zero field using 100 mm samples. The straight sample Jc was
Performance of a 1.29 T Bi-2223 magnet
I (A)
40
80
120
160
200
8
15 T, 0.55% strain
15 T, 0% strain
0.1 T, 0.55% strain
0.1T, 0 % strain
0 T, 0.55% strain
0 T, 0% strain
8
6
E(µV.cm-1)
240
7
6
5
4
4
3
V(µV)
0
10
2
2
1
0
0
-1
-2
0
1
2
3
4
5
J (104 A.cm-2)
6
7
8
9
Figure 2. E–J traces for short samples of Bi-2223 tapes at 0% and 0.55% strain at 4.2 K in fields up to 15 T.
measured as 1.18 × 104 A cm−2 (77 K, 0 T) using a criterion
of 1.5 µV cm−1 . After 10 thermal cycles no degradation was
observed in Jc within the experimental limits. In a second
experiment, the sample was mechanically cycled by bending
it around a 44 mm diameter (the diameter of the inner turn of the
magnet) and then re-straightening. Jc was measured in each
configuration (1 thermal cycle includes two Jc measurements)
during 10 mechanical cycles. The first mechanical cycle
produced a permanent decrease in Jc of ∼20%. Eventually the
Jc degraded by ∼75% from the straight sample performance.
The field dependence of Jc was measured using 30 mm
samples of tape 1 at 4.2 K in fields up to 15 T. Measurements
were taken on a straight sample (0% strain) and a sample bent
on a 44 mm diameter (0.55% strain). Strain values quoted
in this paper were calculated using ε = t/D, where t is
the thickness of the superconductor (0.24 mm) and D is the
bending diameter. Jc of the tape was 8.3 × 104 A cm−2
(4.2 K, 0 T, ε = 0%) and 5.82 × 104 A cm−2 (4.2 K, 0 T,
ε = 0.55%). Data were taken with the field applied in two
and three orthogonal directions with respect to the surface of
the tape for the 0.55 and 0% strained samples, respectively.
Typical E–J characteristics are presented in figure 2 for the
field applied parallel to the tape surface (B parallel to a–b
planes) in fields up to 15 T. At all fields the strain degrades
Jc . However, all traces produce flat baselines to within
the noise (200 nV peak-to-peak) suggesting that undamaged
filaments of zero resistance exist within the strained sample.
In figure 3, the Jc for tape 1 at 4.2 K in fields up to 15 T
with field applied parallel and perpendicular to the tape surface
at 0 and 0.55% strains were determined using a criterion of
1.5 µV cm−1 .
Once the short sample behaviour of the conductor has
been measured the optimum performance of the magnet can
be calculated prior to magnet fabrication assuming that short
sample performance is achieved in long lengths. These
calculations are addressed in section 4.
3. Magnet fabrication
3.1. Magnet windings
The choice of conductor determines whether the R + W, W + R
[6, 17, 31, 32] or wind, react and tighten (WRAT) [18, 33]
methods are most suitable. W + R routes have facilitated the
production of pancake coils with radii ∼5 mm showing no
degradation in Jc [31]. The degradation in tape performance
from the more simple R + W technique due to strain on
the conductor that was used in this work should be weighed
against the high melting point insulation and accurate reaction
conditions for the coils required in the W + R technique. We
calculate that a ∼20% increase in maximum field would have
occurred had a strain-free W + R technique been used.
PIT tape was used in this magnet. There is a continuing
effort to improve processing of PIT tapes, in which precursor
powders are packed into silver alloy tubes, drawn and pressed
into a tape and processed in complex thermomechanical
heat-treatment schedules [5, 34–41].
Other long-length
tapes include dip coating [18, 33, 42] and electrophoretic
deposition [43].
The tape was wound onto Tufnol bobbins of thickness
2 mm and inner diameter 40 mm. Tufnol was chosen primarily
because of its electrically insulating properties and ease of
use but also due to its adequate fracture stress and similar
thermal contraction to epoxy. Copper and other metallic based
materials have been used with the advantages of higher thermal
conductivity and high strength which can be important in large
magnets [7, 15, 16]. Ceramic formers are used in W + R
magnets due to their refractory properties [6, 32].
The bobbins were covered by a single layer of glass
fibre tape and placed on a polytrifluorochloroethylene (PTFE)
mount as shown in figure 4(a). Each coil contained two
oppositely wound pancakes joined together at the bottom
turn using Bi-2223 strips. The tape was wound into anchor
points in the Tufnol bobbin and soldered across the short
orthogonal tapes that provided electrical connections between
435
A B Sneary et al
9
B // surface, perp I, 0% strain
B // surface, // I, 0% strain
B perp surface, perp I, 0% strain
B // surface, 0.55% strain
B perp surface, 0.55% strain
8
200
6
160
5
4
120
3
IC(A)
JC (104 A.cm-2)
7
240
80
2
40
1
0
0
5
10
0
15
B (T)
Figure 3. The critical current density as a function of magnetic field for a short sample of Bi-2223 tape 1 at 0% and 0.55% strain with field
aligned parallel and perpendicular to the tape surface. All data taken at 4.2 K.
Figure 4. (a) The experimental arrangement for winding the tape
and insulation onto the Tufnol bobbin. (b) The PTFE components
used during the vacuum resin impregnation.
the two pancakes. The superconducting tape was co-wound
with non-adhesive polyimide (Kapton) tape of thickness
∼60 µm to provide turn-to-turn insulation. The insulation
was tensioned during winding to maximize the radial packing
factor although the superconductor was not tensioned in order
to minimize damage. Polyimide is usually selected since it is
an insulating material available in thin tape form (∼60 µm)
which has a high (short period) melting point of 310 ◦ C and
reasonable strength (for winding under tension) [7, 14, 16, 44].
436
Alternatives include Mylar, braid or a varnish coating for
R + W [45, 46] and alumina based slurries for W + R coils
[6, 17, 31, 32, 47, 48]. One polyimide sheet was inserted to
provide pancake–pancake insulation. After the first pancake
was wound, the superconducting spool was turned round and
repositioned so that the strain was minimized (i.e. the tape was
not straightened) while winding the second pancake. Finally
a Bi-2223 current lead was soldered to the outer turn of
each pancake and copper voltage taps were soldered onto the
inner and outer turns of each pancake to facilitate diagnostic
measurements on all sections of the coil. Six double-wound
pancake coils were fabricated. The parameters of each coil are
presented in table 2.
The windings of a superconducting magnet system must
be completely impregnated by insulating material which
ensures that the turns do not move when the magnet is
energized. Such movement can quench the magnet and
damage the brittle conductor. Ideally the impregnant should
have a high fracture stress at operating temperature, have
similar thermal contraction properties to the superconductor
and be adhesive to the turns. Standard choices are wax
(often preferred for small coils which are seldom thermally
cycled) and resin (Araldite or Stycast) [46, 49]. The thermal
and mechanical properties vary for different resin systems but
for small magnets the ease of use and pot-life are primary
considerations [50–54]. To key the resin to the bobbins,
dove tail slots were incorporated into the bobbin (cf figure 4).
For the magnet presented here, a Ciba-Geigy resin system
CY1300 (resin), HY906 (hardener) and DY073 (accelerator)
mixed in the ratio 100:80:2 by weight respectively was used.
During impregnation the coils and resin were held in the PTFE
parts which constitute the mould shown in figure 4(b). PTFE
was used because of its non-stick properties. The resin was
degassed for 12 h at 50 ◦ C. The coil was then impregnated with
resin under vacuum at 50 ◦ C. On completion of impregnation,
the system was let up to atmospheric pressure to drive the resin
into the turns. The temperature was then increased to 80 ◦ C
and held for 16 h before curing at 120 ◦ C for 8 h. A minimum
volume of resin was used to avoid any resin-rich areas that are
Performance of a 1.29 T Bi-2223 magnet
Table 2. The main parameters of the constituent coils of the Bi-2223 magnet.
Coil 1 (top)
Coil 2
Coil 3
Pancake
lower
upper
whole
lower
upper
whole
lower
upper
whole
Tape
Height (mm)
Number of turns
Length of tape (m)
Packing factor (%)
1
—
43
8.51
—
1
—
41
8.11
—
—
1
—
43
8.51
—
1
—
43
8.51
—
—
8.0
86
17.02
69.6
1
—
43
8.51
—
2
—
50
9.90
—
—
8.5
93
18.41
65.8
8.5
84
16.62
64.0
Coil 4
Coil 5
Coil 6 (bottom)
Pancake
lower
upper
whole
lower
upper
whole
lower
upper
whole
Tape
Height (mm)
Number of turns
Length of tape (m)
Packing factor (%)
2
—
50
9.90
—
2
—
50
9.90
—
—
8.5
100
17.80
66.0
1
—
43
8.51
—
1
—
43
8.51
—
—
8.5
86
17.02
65.5
1
—
43
8.51
—
1
—
39
7.72
—
—
9.5
82
16.23
55.9
Table 3. Mechanical properties of the constituent materials of the Bi-2223 magnet.
Thermal
contraction at
4.2 K (%)
Yield/fracture
stress (MPa)
a
b
Epoxy resin
[46]
Tufnol (Carp)a
Polyimide [70]
Bi-2223/Ag
[59]
1.1
0.57
0.9
0.25
140
105b
155
—
Manufacturer’s specification.
Room temperature.
particularly liable to crack during thermal cycling. The coil
was removed from the PTFE mould after curing.
In a solenoid the Lorentz forces act to compress the
magnet axially and explode it radially. These forces act to
deform the constituent materials which will modify Jc and
can, if excessive, lead to catastrophic failure. Formulae
for calculating stresses in magnets are well established [46].
If the fracture stress of any of the constituent materials is
exceeded [55], reinforcement should be included such as cowinding thin stainless-steel tapes with the superconductor
[7, 14, 17, 44, 56, 57], reinforcing sections of the magnet [58]
or having a support structure around the outside of the magnet
[46, 58]. Fibre reinforced plastic (FRP) sheets are frequently
used to insulate and reinforce the magnet in the axial direction
[6, 7, 44]. Estimates of the maximum stress produced in our
magnet are around 150 kPa, which is much less than typical
fracture/yield stresses of the constituent materials (listed in
table 3). Typically no additional reinforcement is required for
properly impregnated laboratory-scale magnets.
Coils were stacked and glued together to form the
magnet using the Ciba-Geigy room-temperature curing system
CY1300 (resin) and HY1300 (hardener) mixed in the ratio
3:1 by weight. Joints between the coils were made by soft
soldering Bi-2223 tape connections. Although sophisticated
jointing techniques have been developed, this is rarely
necessary in small magnet systems [32, 49]. The magnet
was supported between two non-magnetic stainless-steel plates
(brass is equally good) using four support rods. A top plate
was connected above the top support plate with 24 voltage
terminals, to allow the voltage across any pancake to be
measured. Current access to all coils was also provided via
Table 4. Dimensions and characteristic properties of the Bi-2223
magnet.
Bore (mm)
Magnet winding I/D (mm)
Magnet winding O/D (mm)
Magnet height (mm)
Number of double pancakes
Length of Bi-2223 tape (m)
Magnet packing factor (%)
Central bore field constant (mT A−1 )
Maximum field constant (mT A−1 )
40
44
82
57.5
6
103.1
58
8.18
8.85
seven current terminals on the top plate electrically connected
to the coils with Bi-2223 tape which allowed any coil to be
individually energized as well as the entire magnet. The
dimensions of the Bi-2223 magnet are presented in table 4
and a photo shown in figure 5.
3.2. External components
Copper current leads were used from the top of the cryostat to
the magnet current terminals. The large heat leak to the system
due to the copper leads was considered acceptable during the
eight day testing period. Optimized permanent current leads
are typically fabricated from tubes of brass with a current
density × length product of 1.5 × 106 A m−1 for operating
at 4.2 K and 6 × 106 A m−1 at 77 K [46, 59, 60].
Quench protection is an essential component in large
superconducting magnets because of the large potential energy
stored when energized (∼1–10 MJ). Critical parameters
include the maximum temperature (θMAX ) and the maximum
437
A B Sneary et al
Figure 5. Photo of Bi-2223 magnet.
voltage (VMAX ) together with the characteristic decay time
for the current decay following a quench. A maximum
temperature of 100 K is normally considered safe enough
to ensure that the local expansion in the region of the (hot)
quench does not damage the windings, although 500 K can
be tolerated in some systems without mechanical damage or
the insulation or resin melting. The breakdown voltage for
polyimide is 120 kV mm−1 and for polymer resins is typically
tens of kV mm−1 . The basic approach to protecting a large
magnet if θMAX or VMAX are too large is to subdivide the
magnet. Choices include active and passive systems either
within or outside the cryostat [46, 61]. Basic calculations
suggest that no sub-division (and protection) is required if the
magnet, as is the case here, consists of less than 2 km of wire
or tape [19].
The inclusion of iron pole pieces [62, 63] or copper coils
[64] has been shown to reduce the magnitude of the radial field
component in the outer coils and therefore increase the field
produced by the magnet. This approach is only applicable for
low field systems where the iron does not saturate. It has been
suggested that the maximum radial component may be reduced
by ∼40% [63]. Calculations suggest that such a reduction may
have produced up to ∼15% increase in maximum field.
4. Performance of coils and magnet
After each coil was fabricated, it was tested at 77 K to identify
any damage or electrical problems. The Ic values obtained
after initial cool-down using the 1.5 µV cm−1 criterion are
presented in table 5. There is relatively large variation from coil
to coil, primarily because the number of turns and geometry of
the coils are not the same. In addition, after the first pancake
of a coil is tested, subsequent measurements on the second
pancake and joints are affected by a remnant field and possibly
because the coil had not cooled back to 77 K. However, since
these ‘quality control’ experiments showed that there was no
evidence of significant damage, all coils were stacked to form
the magnet.
E–J measurements were taken for each coil when
individually energized (to identify any damaged pancakes) at
438
4.2 K and 77 K (not shown), with the whole magnet energized
and with the magnet energized in a background field at 4.2 K
and 77 K. The background field of 296 mT at 4.2 K and 40 mT
at 77 K was provided using a large bore Bi-2223 system which
has been previously reported [65]. Comparing the Ic values
for any pairs of pancakes in a given coil when individually
energized, one finds that for tape 1 the variation in Ic is ∼7%
and for tape 2 it is ∼5%. These variations are consistent with
long-length variations in Ic in the original tape of up to ∼10%.
The E–J characteristics obtained from coil 1 individually
energized at 4.2 K and its component pancakes are presented in
figure 6 and for coil 5 with the whole magnet energized at 4.2
in figure 7. In both cases the baselines are flat (consistent with
the short sample data in figure 2), the superconducting joint
resistance remains low for I < Ic and the E–J characteristics
of the whole coil is the sum of the pancakes and connecting
joint. When the whole magnet is energized, the top and bottom
coils exhibit the lowest Ic at both 4.2 (90 and 92 A) and 77 K
(12.6 and 11.7 A), consistent with other work in the literature
[64–66]. Typical joint resistances between pancakes are below
2 µ at 100 A at 4.2 K. The Ic of the magnet in a self-field
was measured to be 104 A at 4.2 K and 12.8 A at 77 K.
The axial field profile along the central bore axis was
measured using a Hall probe at room temperature with the
magnet energized at 1.0 A. The field profile within the magnet
was calculated to high precision by taking into account the
precise number of turns on each pancake, the packing factor
and the position and geometry of each pancake [67]. In figure 8,
the excellent agreement found between the calculated and
measured profiles provides evidence that there are no shortcircuited turns within the magnet. The measured maximum
field constants of 8.18 mT A−1 at 4.2 K and 8.20 mT A−1 at
room temperature agree to better than 1% with the calculated
value of 8.11 mT A−1 . The axial field is homogeneous to
∼1.5% over 1 cm about the centre.
The optimum shape for small magnets is a (well
documented) compromise between the magnitude of the field
produced, the homogeneity of the magnetic field profile and
the length of conductor [46]. A coil that has a minimum weight
of wire for a given field will produce a homogeneity of ∼1 part
Performance of a 1.29 T Bi-2223 magnet
Table 5. Ic test results of all constituent pancakes and coils of the Bi-2223 magnet at 4.2 and 77 K. At 77 K, each coil was individually
energized and tested before stacking whereas at 4.2 K the equivalent measurement was taken after stacking. The electric field criterion used
for Ic was 1.5 µV cm−1 .
Coil 1 (top)
Pancake
4.2 K
Ic for individually energized coils (A)
n-index
Ic for magnet energized (A)
Ic for whole magnet energized in
296 mT background field (A)
77 K
Ic for individually energized coils (A)
n-index
Ic for whole magnet energized (A)
Ic for whole magnet energized in 296 mT
background field (A)
Coil 2
lower
upper
whole
lower
upper
whole
lower
upper
whole
115
14.7
99
123
20.0
90
124
—
95
120
16.7
112
127
16.7
110
127
—
111
115
16.1
—
103
14.9
110
111
—
114
96
88
95
109
109
109
119
106
113
18.3
9.7
13.5
14.9
9.0
12.6
15.8
8.6
13.3
20.1
10.5
16.0
16.4
10.2
15.2
17.2
9.2
15.8
19.1
10.5
17.9
13.7
10.3
15.7
14.8
9.8
16.5
12.7
12.2
12.6
15.3
14.6
14.8
16.7
14.9
15.4
Coil 4
Pancake
4.2 K
Ic for individually energized coils (A)
n-index
Ic for magnet energized (A)
Ic for whole magnet energized in
296 mT background field (A)
77 K
Ic for individually energized coils (A)
n-index
Ic for whole magnet energized (A)
Ic for magnet energized in 40 mT
background field (A)
Coil 3
Coil 5
Coil 6 (bottom)
lower
upper
whole
lower
upper
whole
lower
upper
whole
104
15.2
107
108
17.3
117
105
—
110
130
16.3
107
128
15.9
115
130
—
111
123
15.1
92
120
15.3
100
—
—
94
107
117
112
108
115
109
88
97
95
17.7
11.5
15.3
14.7
11.3
17.6
15.3
10.9
15.8
19.2
10.3
14.7
16.0
10.3
16.1
16.9
9.6
15.9
17.5
9.8
11.7
15.3
9.1
14.0
16.0
9.1
12.5
14.6
16.3
15.0
14.0
15.5
14.7
11.3
13.0
12.1
I (A)
0
40
60
80
100
120
Inner pancake
Whole coil
Outer pancake
Joint
Inner pancake Ic criterion
Outer pancake Ic criterion
2
V (mV)
20
1
0
0
1
2
3
4
4
-2
J (10 A.cm )
Figure 6. Electric field–current density characteristics of the components of coil 1 with the coil individually energized at 4.2 K.
in 103 over a 1 cm diameter sphere volume. For higher
homogeneity up to 1 part in 105 , notch, compensation or shim
coils may be added [19, 68]. The maximum magnet constant
for our magnet is calculated as 8.85 mT A−1 , positioned on
the inner winding at 2 mm above the centre. This maximum
is offset from the centre due to the asymmetry of the number
of turns. The maximum current passed through the magnet
in a self-field was 146 A at 4.2 K, limited by a 10 V power
439
A B Sneary et al
I (A)
0
2.0
20
40
60
80
100
120
Inner pancake
Whole coil
Outer pancake
Joint
Ic criterion for pancakes
1.0
1
V(mV)
E (µV.cm-1)
1.5
0.5
0.0
0
-0.5
0
1
2
4
3
4
-2
J (10 A.cm )
Figure 7. Electric field–current density characteristics of the components of coil 5 with the whole magnet energized at 4.2 K.
9
Measured field
Calculated field
Field constant (mT.A-1)
8
7
6
5
4
Bottom
-30
Top
-20
-10
0
10
20
Distance from centre of magnet (mm)
30
Figure 8. Comparison between measured and calculated axial field profiles within the Bi-2223 magnet.
supply, corresponding to a maximum bore field of 1.19 T and
a maximum magnet field on the turns of 1.29 T. In a background
field at 4.2 K the maximum measured bore field was 1.470 T
(including 296 mT contribution from the large bore magnet).
5. Calculations of magnet performance
The Ic of any turn in any pancake is determined by the local
angle and magnitude of the magnetic field and the local strain
state. The magnitude and direction of field throughout the
magnet has been calculated using standard formulae. The
field profile of the magnet was generated using the precise
dimensions of each coil. The geometry and number of turns of
coil 2 were used in calculations for the individually energized
coils. The strain state was calculated from the bending radius.
440
It is a very large experimental task to determine the
magnetic field, temperature and strain dependence of Jc for
a short sample throughout the superconducting phase. To
approximate the functional form of Jc over the required range,
we have chosen to linearly interpolate the variable magnetic
field Jc data in figure 3 obtained at two different orientations
of the field and two different strains. Comparison with the
literature shows that these interpolations give a reasonable
first approximation to the observed Bi-2223 Jc dependence
on strain [27–30] and angle [20–26] as discussed in section 6.
The positions of the turns with the lowest Ic (i.e. the
performance limiting turn) were identified at r = 36 mm
(105 A) for the individually energized coils and on the top
and bottom pancake at r = 38 mm (94 A) for the fully
energized magnet. The Jc dependences and corresponding
Performance of a 1.29 T Bi-2223 magnet
8
240
Predicted Jc(B) for magnet at r = 38 mm
Magnet load line
Predicted Jc(B) for coil at r = 36 mm
Coil load line
7
200
160
5
4
120
IC(A)
JC (104 A.cm-2)
6
Ic = 105 A
3
80
Ic = 94 A
2
40
1
0
0.0
0.1
0.2
0.3
0.4
0.5
B (T)
0.6
0.7
0.8
0.9
0
1.0
Figure 9. The field dependence of the critical current density calculated for the turn at r = 36 mm in an individually energized coil and for
the turn at r = 38 mm on the outermost pancake with the magnet energized. The associated load lines are also shown.
70
60
0.9
50
0.8
40
0.7
30
20
Normalized Ic
Normalized field
Angle
0.6
10
0.5
Bore
Outer turn
0.4
0.30
0.35
0.40
0.45
0.50
Angle between field and c-axis (o)
Normalized Ic (Ic/Icmax) and B(B/Bmax)
1.0
0
0.55
Strain (%)
Figure 10. The change in the normalized critical current (•), normalized magnetic field (
) and angle of field () from the c-axis as a
function of strain for turns across the outermost pancake with the magnet energized. Ic is normalized to the value of the outer turn and the
field is normalized to the field experienced by the inner turn.
load lines for the performance limiting turns are presented in
figure 9. The change in the normalized current density within
the top pancake when the magnet is energized is presented in
figure 10 where calculations for Jc were made at 2 mm intervals
radially along the pancake. The changes in magnitude and
direction of the field experienced by the turns are presented in
the same figure. On moving radially outwards from the bore the
magnitude of the field decreases by 50%, the strain decreases
by 45% and the angle of the field becomes more closely aligned
with the c-axis of the tape by 60◦ . The combined effect of these
factors results in only a 10% variation in Ic across the pancake.
The calculated Ic of the top pancake is 94 A. Above 94 A,
heat will start to be generated locally. An average value of
calculated using a length
Ic (av) for the pancakes has also
been
average defined by Ic (av) = Ic l/ l where l is a length of
tape with a particular Ic . For an individually energized coil,
Ic (av) is 115 A and 101 A across the top coil with the magnet
energized. The measured Ic values of 90 and 92 A for the top
and bottom pancakes with the magnet energized (cf table 5)
are in good agreement with the calculated value of 94 A.
The performance of the magnet has also been calculated
assuming a W + R approach. Necessarily, there is no
degradation in the tape performance due to strain. In the case
of the individually energized coil, the performance limiting
turn is positioned at r = 36 mm with an Ic value of 133 A.
With the magnet energized, the performance limiting turn is
positioned at r = 38 mm on the outermost pancake with an
Ic of 117 A. Hence the maximum field could be increased by
∼20%.
6. Discussion
We estimate the uncertainty in the predicted Jc due to the
anisotropy and field interpolation to be less than ∼5% given
441
A B Sneary et al
that the field at the performance limiting turn is aligned with
the c-axis to ∼5◦ (cf figure 10). The interpolation to determine
the strain dependence of Jc within the magnet is more complex.
Our experience measuring the strain dependence of Jc in brittle
Bi-2223 tapes is that there is considerable statistical variation
in short sample measurements [69], probably because of the
variation in porosity of the tapes along their length [10]. It is
therefore difficult to obtain variable strain short sample data
that are representative of long lengths. Nevertheless, in light
of the ∼10% variation in Ic over long lengths in the tapes
and the ∼20% degradation in Ic due to strain intrinsic to the
R + W technique, there is very good agreement between the
calculations and the final magnet performance. We conclude
that a reliable fabrication process has been developed for
laboratory-scale magnets which produces minimal additional
damage due to handling.
We suggest fabrication of a laboratory-scale magnet as
follows; measure Jc of the tape as a function of field with the
field applied parallel and orthogonal to the tape surface in the
unstrained state and at the strain state equal to that on the bore
of the proposed magnet; measure Jc after many thermal cycles
to ensure no degradation occurs; calculate the performance of
the proposed magnet; if necessary a third interpolation point
can be obtained on a short sample as a function of the field
at the strain state and orientation of the magnetic field for the
turn with the lowest Ic , and the expected performance of the
magnet double-checked; construct the coils; test the coils in a
self-field at 77 K and replace any damaged ones; finally stack
the coils to form the magnet.
7. Conclusions
In this paper, the design and fabrication of a 1.29 T laboratorysized Bi-2223 magnet operating at 4.2 K has been described.
The magnet was fabricated from six resin impregnated doublewound pancakes of inner diameter 40 mm wound via the
R+W route using Bi-2223/Ag tape with critical current density
8.3 × 104 A cm−2 (4.2 K, 0 T). Design considerations actually
used and other possible options in fabricating a similar-sized
magnet with different operational requirements or restrictions
have been discussed.
The measured performance of the magnet and all
constituent coils has been compared with calculated values
obtained using field profile calculations and short sample Jc
data. Good agreement between calculation and experiment
has been found which shows that the top and bottom pancakes
limit the performance of the magnet. The performance
of the magnet is reduced by ∼20% due to the strain on
the superconductor produced during the R + W technique.
We conclude that a reliable fabrication technique has been
described for the non-specialist to produce a R + W hightemperature superconducting laboratory-scale magnet.
Acknowledgments
The authors wish to thank H Jones and P Richens in the Magnet
Group at Oxford University for their help and advice on coil
winding, R Prowse for technical support winding the magnets
at BICC General, G Teasdale and P Armstrong for fabrication
of the magnet component parts, and P Russell and V Greener
442
in help producing figures 4 and 5. This work was supported
by EPSRC and BICC General.
References
[1] Bednorz J and Muller K 1986 Z. Phys. B 64 189–93
[2] Wu M, Ashburn J, Torng C, Hor P, Meng R, Gao L, Huang Z,
Wang Y and Chu C 1987 Phys. Rev. Lett. 58 908–10
[3] Li Q, Brodersen K, Hjuler H and Freltoft T 1993 Physica C
217 360–6
[4] Malozemoff A et al 1999 IEEE Trans. Appl. Supercond. 9
2469–74
[5] Balachandran U, Iyer A, Jammy R, Chudzik M, Lelovic M,
Krishnaraj P, Eror N and Haldar P 1997 IEEE Trans. Appl.
Supercond. 7 2207–10
[6] Zeng R, Lui H, Beales T and Dou S 1999 IEEE Trans. Appl.
Supercond. 9 2605–8
[7] Ohkura K, Sato K, Ueyama M, Fujikami J and Iwasa Y 1995
Appl. Phys. Lett. 67 1923–5
[8] Kaneko T, Hikata T U M, Mikumo A, Ayai N, Kobayashi S,
Saga N, Hayashi K, Ohmatsu K and Sato K 1999 IEEE
Trans. Appl. Supercond. 9 2465–8
[9] Leghissa M, Fischer B, Roas B, Jenovelis A, Wiezoreck J,
Kautz S, Neumuller H, Reimann C, Nanke R and Muller P
1997 IEEE Trans. Appl. Supercond. 7 355–8
[10] Larbalestier D and Lee P 1999 Proc. 1999 Particle Accelerator
Conf. pp 177–81
[11] Sneary A, Friend C, Vallier J and Hampshire D 1999 IEEE
Trans. Appl. Supercond. 9 2585–8
[12] Sato K, Hikata T and Iwasa Y 1990 Appl. Phys. Lett. 57
1928–9
[13] Snitchler G, Kaksi S, Manlief M, Schwall R, Sid-Yekhlef A,
Ige S, Medeiros R, Francavilla T and Gubser D 1999 IEEE
Trans. Appl. Supercond. 9 553–8
[14] Sato K, Kato T, Ohkura K, Kobayashi S, Fujino K, Ohmatsu K
and Hayashi K 2000 Supercond. Sci. Technol. 13 18–22
[15] Zeng R, Zhou Y, Hua P, Fu X, Ye B, Zhou M, Yuan G, Lui H
and Dou S 1998 Supercond. Sci. Technol. 11 535–9
[16] Kumakura H, Kitaguchi H, Togano K, Wada H, Ohkura K,
Ueyama M, Hayashi K and Sato K 1998 Cryogenics 38
639–43
[17] Verges P, Fischer K, Hutten A and Fuchs G 1998 Cryogenics
38 607–11
[18] Tomita N, Arai M, Yanagisawa E, Morimoto T, Kitaguchi H,
Kumakura H, Togano K and Nomura K 1996 Cryogenics 36
485–90
[19] Kerley N 1998 Handbook of Applied Superconductivity; Vol. 2
ed B Seeber (Bristol: IOP Publishing) pp 1067–89
[20] Rabara M, Takeuchi T and Miya K 1999 Physica C 313
213–18
[21] Oota A and Tanaka M 1996 Physica C 268 295–9
[22] Fuller-Mora W 1996 Phys. Rev. B 54 11 977–80
[23] Han G, Han H, Wang Z, Wang S, Wang F, Yuan W and Chen J
1994 Cryogenics 34 613–16
[24] Hensel B, Grivel J, Jeremie A, Lerin A, Pollini A and
Flukiger R 1993 Physica C 205 329–37
[25] Hu Q, Weber H, Lui H, Dou S and Neumuller H 1995 Physica
C 252 211–20
[26] Kovak P, Husek I, Rosova A and Pachla W 1999 Physica C
312 179–90
[27] Ekin J, Finnemore D, Li Q, Tenbrink J and Carter W 1992
Appl. Phys. Lett. 61 858–60
[28] Polak M, Parrell J, Polyanskii A, Pashitski A and
Larbalestier D 1997 Appl. Phys. Lett. 70 1034–6
[29] Suenga M, Fukumoto Y, Haldar P, Thurston T and
Wildgruber U 1995 Appl. Phys. Lett. 67 3025–7
[30] Huang Y, ten Haken B and ten Kate H 1999 IEEE Trans. Appl.
Supercond. 9 2702–5
[31] Kovac P, Cesnak L, Melisek T, Husek I, Bukva P, Pitel J,
Kopera L, Pachla W and Bucholtz W 1999 Supercond. Sci.
Technol. 12 507–13
Performance of a 1.29 T Bi-2223 magnet
[32] Vo N 1998 J. Magn. Magn. Mater. 188 145–52
[33] Newson M, Ryan D, Wilson M and Jones H 2000 IEEE Trans.
Appl. Supercond. 10 468–71
[34] Hellstrom E 1995 High-temperature Superconducting
Materials Science and Engineering ed D Shi (New York:
Elsevier) pp 383–440
[35] Dou S and Liu H 1993 Supercond. Sci. Technol. 6 297–314
[36] Flukiger R, Graf T, Decroux M, Groth C and Yamada Y 1991
IEEE Trans. Appl. Supercond. 27 1258–63
[37] Grasso G, Jeremie A and Flukiger R 1995 Supercond. Sci.
Technol. 8 827–32
[38] Marti F, Grasso G, Grivel J-C and Flukiger R 1998 Supercond.
Sci. Technol. 11 485–95
[39] Zeng R, Beales T, Lui H and Dou S 1998 Supercond. Sci.
Technol. 11 299–303
[40] Han Z, Skov-Hansen P and Freltoft T 1997 Supercond. Sci.
Technol. 10 371–87
[41] Vo N, Willis J, Peterson D, Liu H and Dou S 1998 Physica C
299 315–26
[42] Picard J et al 1999 IEEE Trans. Appl. Supercond. 9 535–40
[43] Yang M, Goringe M, Grovenor C, Jenkins R and Jones H 1994
Supercond. Sci. Technol. 7 378–88
[44] Ohmatsu K, Hahakura S, Kato T, Fujino T, Ohkura K and
Sato K 1999 IEEE Trans. Appl. Supercond. 9 924–7
[45] Tupper M, Fabian P and Beavers F 2000 IEEE Trans. Appl.
Supercond. 10 1350–3
[46] Wilson M 1998 Superconducting Magnets (New York: Oxford
University Press)
[47] Celic E, Mutlu I, Avci E and Hascicek Y 2000 IEEE Trans.
Appl. Supercond. 10 1329–32
[48] Celic E, Mutlu I and Hascicek Y 2000 IEEE Trans. Appl.
Supercond. 10 1341–4
[49] Collings E 1986 Applied Superconductivity, Metallurgy and
Physics of Titanium Alloys vol 2 (New York: Plenum)
[50] Evans D and Morgan J 1986 Cryogenic Engineering
ed B Hands (London: Academic) pp 271–92
[51] Lee H and Neville K 1982 Handbook of Epoxy Resins (New
York: McGraw-Hill)
[52] Baldan C, Shigue C and Filho E 2000 IEEE Trans. Appl.
Supercond. 10 1347–9
[53] Jones H and Hickman A 2000 IEEE Trans. Appl. Supercond.
10 1345–6
[54] Brennan A, Miller T, Arnold J, Huang K, Gephart N and
Markewicz W 1995 Cryogenics 35 783–5
[55] Hands B 1986 Cryogenic Engineering ed B Hands (London:
Academic) pp 241–70
[56] Sakuraba J, Mikami Y, Watazawa K, Watanabe K and Awaji S
2000 Supercond. Sci. Technol. 13 12–17
[57] Kitaguchi H, Kumakura H, Togano K, Okada M, Tanaka K
and Sato J 1998 Cryogenics 38 181–6
[58] Jones H, Van Cleemput M, Hickman A, Ryan D and Saleh P
1998 Physica B 246–247 337–40
[59] Keys S and Hampshire D 2001 Handbook of Applied
Superconductivity (Bristol: IOP Publishing) to be published
[60] Herrmann P 1998 Handbook of Applied Superconductivity;
Vol. 1 ed B Seeber (Bristol: IOP Publishing) pp 801–43
[61] Meb K 1998 Handbook of Applied Superconductivity; Vol. 1
ed B Seeber (Bristol: IOP Publishing) pp 527–55
[62] Selvaggi J, Mathewson W, Laughman R, Francavilla T,
Gubser D and Miller M 1996 Cryogenics 36 555–8
[63] Fabbricatore P, Priano C, Testa M, Musenich R, Kovac P,
Martone A, Petrillo E and Ariante M 1998 Supercond. Sci.
Technol. 11 304–10
[64] Pitel J and Kovac P 1997 Supercond. Sci. Technol. 10
847–52
[65] Sneary A, Friend C, Richens P, Jones H and Hampshire D
1999 IEEE Trans. Appl. Supercond. 9 936–9
[66] Daumling M and Flukiger R 1995 Cryogenics 35 867–70
[67] Montgomery D 1969 Solenoid Magnet Design (New York:
Wiley)
[68] Tschopp W and Laukien D 1998 Handbook of Applied
Superconductivity; Vol 2 ed B Seeber (London: IOP)
p 1191–212
[69] Hamid H and Hampshire D 1998 Cryogenics 38 1007–15
[70] Gerhold J 1998 Handbook of Applied Superconductivity;
Vol. 1 ed B Seeber (Bristol: IOP Publishing) pp 1121–37
443